How to right-size your desk ratio with booking data

Almost every organisation that moves to hybrid work asks the same question within a year: how many desks do we actually need? The honest answer is fewer than you have and more than your average suggests. Getting from that answer to a number you can put in front of a finance committee takes data you probably already collect.

Why badge data misleads

Access badges record entries, not desk use. They count the person who came in for a two-hour meeting and never sat down, miss the colleague who tailgated through the door, and say nothing about where in the building people went. They are useful for security and rough attendance trends, but they are the wrong tool for sizing a floor.

Desk bookings with check-in are better, because they record what you are actually trying to provide: a desk, on a day, in a zone, used by someone. The check-in is what makes them honest. Without it, bookings measure intentions; with it, they measure use.

The four numbers that matter

• Peak occupancy: the highest number of desks checked in at the same time, per day.
• Average occupancy: the mean across the week, useful for energy and services but not for sizing.
• No-show rate: bookings that were never checked in. A high rate means your floor looks fuller than it is.
• Occupancy by zone: teams cluster, and a building can be full on one floor and empty on another.

Size for the peak, not the average. An office that is 55% occupied on average may still be completely full on Tuesdays, and nobody remembers the average when they cannot find a seat.

Use a high percentile, not the maximum

Sizing for the single busiest day of the year — the all-hands, the week before the holidays — builds a building for one day. Instead, take the daily peaks over at least eight to twelve weeks and use a high percentile. The 85th percentile is a common choice: on most days everyone finds a desk, and on the few busiest days people sit in project rooms, lounges or a neighbouring zone.

Then add a buffer for things your booking data cannot see yet: headcount growth, visitors and contractors who need a desk, and the natural friction of people wanting to sit near their team. Ten to fifteen percent is typical.

A worked example

Take an illustrative office with 400 people and 300 desks. Over twelve weeks, check-ins peak on Tuesdays and Thursdays and fall away on Fridays. The 85th percentile of the daily peak is 212 desks. With a 12% buffer, the office needs about 237 desks — a ratio of roughly 0.6 desks per person, and around 60 desks, or one small floor, more than necessary.

Look at zones before you act. If the product and engineering floors run at 95% on Tuesdays while the finance floor sits at 40%, the first move is not to remove desks. It is to redraw neighbourhoods so that the teams who come in on the same days sit on the same floors.

Shape the week before you cut

Peaks are partly a policy choice. When every team picks Tuesday and Thursday as anchor days, you build for a two-day week. Staggering anchor days by team — some on Monday and Wednesday, others on Tuesday and Thursday — flattens the peak and can reduce the desks you need by more than any ratio calculation.

Fix no-shows at the same time. If one booking in ten is never used, automatic release after a grace period gives those desks back to the people standing in the corridor, and your peak numbers become trustworthy.

Revisit every quarter

Hybrid patterns keep moving: new joiners come in more often, projects bring teams in for a few months, seasons change commuting habits. Recalculate the ratio every quarter from the latest twelve weeks and make small, reversible changes — closing a floor on Fridays, converting a row of desks into a project room — before you make large ones. A lease decision based on a year of honest data is much easier to defend than one based on a feeling that the office looks quiet.